Nice idea Mr Grumble but in reality you cannot start feeding the ball in that soon or closing the bolt so soon on a half fed ball. It's because of the actual shape of a sphere it's very limiting in terms of how far in it needs to be fed before you can start closing on it...
I calculated it out years ago but to be honest can't remember my figures for how soon you could close the bolt without having a serious risk of chopping the ball... It made only a very small difference.
I'll let some pictures speak for themselves...
Here you can see a ball at when it starts to fall and a ball half way into the breech. I think we can definitely say with a ball half way in this would be the worst point possible for the bolt to start coming forwards.
Here you can see a ball 3/4 of the way in. You can see even now if the bolt starts moving it is very likely to pinch/chop the ball.
You can see when a ball is fully chambered that the closest point to the front of the breech is not actually very far, this is due to the shape of a sphere.
Below you can see that with the ball almost fed all the way in the bolt is still only able to come forwards 3mm without starting to pinch and chop the ball...
Here you can see it from the front and see just how likely that ball is to be chopped.
Here you can how the second ball in a stack can drop slightly into the breech and then git hit by the bolt. This is likely the cause of more broken paint than you would think as the bolt flies forwards and cracks the second ball. This is also appearing as a problem with the Halo as it sometimes forces a second ball slightly into the breech and thus it gets broken by the bolt firing the prior ball. This is an issue due to large bore breeches and bad ball detents.
Another view of that second ball getting smacked as the first is fired.
In all of my testing of vert I've found that the feed rates do actually tie in remarkably well with the theoretical rates that the maths shows us, giving that the ball has to be fully chambered before the bolt comes forwards. The only differences tending to be when blowback and bobble reduces how well it can feed.
I don't have the same testing facilities as AGD but I have done quite a bit of research into this stuff Some things I said about bolt speed back in '97 are just starting to be proven true.